Biogeosciences, 14, 4619–4635, 2017https://doi.org/10.5194/bg-14-4619-2017© Author(s) 2017. This work is distributed underthe Creative Commons Attribution 3.0 License.

Expansion of oil palm and other cash crops causes an increase of theland surface temperature in the Jambi province in IndonesiaClifton R. Sabajo1,2, Guerric le Maire3, Tania June4, Ana Meijide1, Olivier Roupsard3,5, and Alexander Knohl1,6

1University of Göttingen, Bioclimatology, 37077 Göttingen, Germany2AgroParisTech – Centre de Montpellier, Agropolis International, 648 rue Jean-François Breton, 34093 Montpellier, France3CIRAD, UMR Eco&Sols, 34398 Montpellier, France4Agrometeorology Laboratory Department of Geophysics and Meteorology, Faculty of Mathematics and Natural Sciences,Bogor Agricultural University (IPB), Bogor, Indonesia5CATIE, Centro Agronómico Tropical de Investigación y Enseñanza/Tropical Agriculture Centre for Research and HigherEducation, 7170 Turrialba, Costa Rica6University of Göttingen, Centre of Biodiversity and Sustainable Land Use (CBL), 37073 Göttingen, Germany

Correspondence to: Clifton R. Sabajo (csabajo@uni-goettingen.de)

Received: 21 May 2017 – Discussion started: 29 May 2017Revised: 28 September 2017 – Accepted: 29 September 2017 – Published: 25 October 2017

Abstract. Indonesia is currently one of the regions with thehighest transformation rate of land surface worldwide re-lated to the expansion of oil palm plantations and other cashcrops replacing forests on large scales. Land cover changes,which modify land surface properties, have a direct effecton the land surface temperature (LST), a key driver formany ecological functions. Despite the large historic landtransformation in Indonesia toward oil palm and other cashcrops and governmental plans for future expansion, this isthe first study so far to quantify the impacts of land trans-formation on the LST in Indonesia. We analyze LST fromthe thermal band of a Landsat image and produce a high-resolution surface temperature map (30 m) for the lowlandsof the Jambi province in Sumatra (Indonesia), a region whichsuffered large land transformation towards oil palm and othercash crops over the past decades. The comparison of LST,albedo, normalized differenced vegetation index (NDVI) andevapotranspiration (ET) between seven different land covertypes (forest, urban areas, clear-cut land, young and matureoil palm plantations, acacia and rubber plantations) showsthat forests have lower surface temperatures than the otherland cover types, indicating a local warming effect after for-est conversion. LST differences were up to 10.1± 2.6 ◦C(mean±SD) between forest and clear-cut land. The differ-ences in surface temperatures are explained by an evapora-tive cooling effect, which offsets the albedo warming effect.

Our analysis of the LST trend of the past 16 years based onMODIS data shows that the average daytime surface tem-perature in the Jambi province increased by 1.05 ◦C, whichfollowed the trend of observed land cover changes and ex-ceeded the effects of climate warming. This study providesevidence that the expansion of oil palm plantations and othercash crops leads to changes in biophysical variables, warm-ing the land surface and thus enhancing the increase of theair temperature because of climate change.

1 Introduction

Indonesia is one of the regions where the expansion of cashcrop monocultures such as acacia (timber plantations), rub-ber, oil palm plantations and smallholder agriculture hasdrastically reduced the area of primary forest in the last 2.5decades (Bridhikitti and Overcamp, 2012; Drescher et al.,2016; Marlier et al., 2015; Miettinen et al., 2012; Verstraetenet al., 2005). This large-scale conversion of rainforest foragricultural use has been observed on the island of Sumatra,which has experienced the highest primary rainforest coverloss in all of Indonesia (Drescher et al., 2016; Margono et al.,2012; Miettinen et al., 2011). Forest cover in the Sumatranprovinces of Riau, North Sumatra and Jambi declined from93 to 38 % of provincial area between 1977 and 2009 (Mietti-

Published by Copernicus Publications on behalf of the European Geosciences Union.

4620 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

nen et al., 2012). These large-scale transformations, observedas land cover change, and land use intensification have led tosubstantial losses in animal and plant diversity, ecosystemfunctions and changed microclimatic conditions (Clough etal., 2016; Dislich et al., 2016; Drescher et al., 2016). Ad-ditionally, these changes directly alter vegetation cover andstructure and land surface properties such as albedo, emis-sivity and surface roughness, which affect gas and energyexchange processes between the land surface and the atmo-sphere (Bright et al., 2015).

Replacing natural vegetation with another land cover mod-ifies the surface albedo, which affects the amount of solarradiation that is absorbed or reflected and consequently al-ters net radiation and local surface energy balance. A loweror higher albedo results in a smaller or greater reflection ofshortwave radiation. As a result, the higher or lower amountsof net radiation absorption may increase or decrease the sur-face temperature and change evapotranspiration (ET) (Mah-mood et al., 2014).

Changes in land cover also alter surface emissivity, i.e. theratio of radiation emitted from a surface to the radiation emit-ted from an ideal black body at the same temperature follow-ing the Stefan–Boltzmann law. Emissivity of vegetated sur-faces varies with plant species, density, growth stage, watercontent and surface roughness (Snyder et al., 1998; Weng etal., 2004). A change of emissivity affects the net radiation be-cause it determines the emission of longwave radiation thatcontributes to radiative cooling (Mahmood et al., 2014).

Water availability, surface type, soil humidity, local atmo-spheric and surface conditions affect the energy partition-ing into latent (LE), sensible (H) and ground heat (G) fluxes(Mildrexler et al., 2011). Surface roughness affects the trans-ferred sensible and latent heat by regulating vertical mixingof air in the surface layer (van Leeuwen et al., 2011), therebyregulating land surface temperature (LST). Through its asso-ciation with microclimate, net radiation and energy exchange(Coll et al., 2009; Sobrino et al., 2006; Voogt and Oke, 1998;Weng, 2009; Zhou and Wang, 2011), LST is a major land sur-face parameter, and as a climatic factor it is regarded to be amain driver of diversity gradients related to the positive re-lationships between temperature and species richness (Wanget al., 2016).

The replacement of natural vegetation also changes ET(Boisier et al., 2014) and LST because the surface biophys-ical variables (i.e. surface albedo, LST, emissivity and indi-rectly leaf area index (LAI) and normalized difference vege-tation index (NDVI)) are interconnected through the surfaceradiation balance. When ET decreases, for example, surfacetemperatures and sensible heat fluxes increase; in contrast,when ET increases, the increased LE fluxes lower surfacetemperatures and decrease H fluxes (Mahmood et al., 2014)under equal net radiation conditions because with a changein vegetation, soil and ecosystem heat flux and net radiationalso change due to an alteration of the biophysical variables.Vegetation structure, represented by NDVI, LAI and vegeta-

tion height, is in this respect an important determinant of theresistances or conductivities to heat, moisture and momen-tum transfer between the canopy and the atmosphere (Brightet al., 2015), facilitating the amounts/ratios of sensible heatto water vapor dissipation away from the surface (Hoffmannand Jackson, 2000).

To understand the effects of land cover changes on LST,the associated biophysical variables must be evaluated. Thiscan be done through the surface radiation budget and energypartitioning which unite these biophysical variables directlyor indirectly: albedo as direct determinant of the net solarradiation, NDVI as a vegetation parameter determining theemissivity, which in turn determines the amount of reflectedand emitted longwave radiation; LST directly affecting theamount of emitted longwave radiation from the surface; andET, which affects the amount of energy that is used for sur-face cooling via the evaporation of water.

The effect of land cover change on LST is dependent onthe scale, location, direction and type of the change (Lon-gobardi et al., 2016). Several studies showed an LST in-crease after forest conversion to built-up areas, agriculturalland (Zhou and Wang, 2011), crop land and pasture lands(Peng et al., 2014) in China. Similar observations were re-ported for South American ecosystems: low vegetation suchas grasslands in Argentina were warmer than tall tree vege-tation (Nosetto et al., 2005). In Brazil, the surface tempera-ture increased after the conversion of natural Cerrado vege-tation (a savanna ecosystem) into crop/pasture (Loarie et al.,2011a). Similar effects were also observed for other SouthAmerican biomes (Salazar et al., 2016). In a global analysis,Li et al. (2015) showed that the cooling of forests is moder-ate at midlatitudes but northern boreal forests are warmer, anindication that the effect of land cover change on LST varieswith the location of the land cover change (Longobardi et al.,2016). Similar studies on the Indonesian islands are lackingbut surface temperature increases are expected as an effect ofoil palm and cash crop land expansion in the recent decades.

Measuring LST changes is critical for understanding theeffects of land cover changes, but challenging. LST can bemonitored with LST products retrieved from thermal in-frared (TIR) remote sensing data: e.g. the use of the ther-mal bands of the Moderate Resolution Imaging Spectrom-eter (MODIS) on board the Terra and Aqua satellite (So-brino et al., 2008), the thermal band of the Thematic Map-per (TM) on board the LANDSAT-5 platform (Sobrino etal., 2004, 2008) or Enhanced Thematic Mapper (ETM+) onboard the LANDSAT-7 platform. The advantage of MODISdata is the availability of readily processed products at hightemporal resolution (daily) at medium (250–500 m) to coarse(1000–5000 m) spatial resolution scale; MODIS LST prod-uct (MOD11A1/MYD11A1), for example, is provided at adaily temporal resolution with a spatial resolution of 1 km.Landsat data are provided at a higher spatial resolution(30 m), but the temporal resolution is limited to 16 days andthe retrieval of LST requires the correction of the satellite-

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4621

observed radiances for atmospheric absorption and emission(Coll et al., 2009). Besides LST, the connected biophysi-cal variables of the energy and radiation budget can be de-rived from the visible and near-infrared (VIS–NIR) bands ofMODIS or Landsat, making integrated monitoring of the bio-physical variables related to changing land surface possible.In Indonesia, a large proportion of the land use changes isdriven by smallholders (Dislich et al., 2016), and thus a com-bination of Landsat (for a fine spatial resolution) and MODIS(for temporal developments) seems desirable.

The modification of the physical land surface propertiesinfluences climate and local microclimatic conditions viabiogeochemical and biophysical processes. Therefore, givenIndonesia’s history of large-scale agricultural land conver-sion and governmental plans to substantially expand the oilpalm production (Wicke et al., 2011), it is important to studythe effects of the expansion of cash crop areas on the bio-physical environment, especially on LST as a key land sur-face parameter. These effects have been poorly studied in thisregion and, according to our knowledge, this is the first studyto quantify the effects of land use change on LST in Indone-sia. We focus on the Jambi province (in Sumatra, Indonesia)as it experienced large land transformation towards oil palmand other cash crops such as rubber plantations in the past,and it may serve as an example of future changes in otherregions.

Our main objective is to quantify the differences in LSTacross different land cover types and to assess the impact ofcash crop expansion on the surface temperature in the Jambiprovince in the past decades (2000–2015). With this study weaim (1) to evaluate the use of Landsat and MODIS satellitedata as sources of a reliable surface temperature estimation ina tropical region with limited satellite data coverage by com-paring the surface temperatures retrieved from both satel-lite sources to each other and against ground observations,(2) to quantify the LST variability across different land covertypes, (3) to assess the long-term effects of land transforma-tion on the surface temperature against the background ofclimatic changes and (4) to identify the mechanisms that ex-plain the surface temperature changes caused by alterationsof biophysical variables. In this study we compare the sur-face temperatures of different land cover types that replaceforests (i.e. oil palm, rubber and acacia plantations, clear-cutland and urban areas) by using high-resolution Landsat andmedium-resolution MODIS satellite data and discuss the dif-ferences by taking into account other biophysical variablessuch as the albedo, NDVI and ET.

2 Materials and methods

2.1 Study area

The study was carried out in the lowlands (ap-prox. 25 000 km2) of the Jambi province (total area

50 160 km2) on Sumatra, Indonesia, between latitudes 0◦30′

and 2◦30′ S and longitudes 101◦ and 104◦30′ E (Fig. 1).This region has undergone large land transformation towardsoil palm and rubber plantations over the past decades andthus may serve as an example of expected changes in otherregions of Indonesia (Drescher et al., 2016). The area hasa humid tropical climate with a mean annual temperatureof 26.7± 0.2 ◦C (1991–2011, annual mean±SD of theannual mean), with little intra-annual variation. Mean annualprecipitation was 2235± 381 mm and a dry season withless than 120 mm monthly precipitation usually occurredbetween June and September (Drescher et al., 2016). Pre-viously logged rainforests in the Jambi province have beenconverted to intensively managed agro-industrial productionzones and into smallholder farms to grow cash crops ofrubber (Hevea brasiliensis) and oil palm (Elaeis guineensis)or fast-growing tree species such as Acacia mangium forpulp production (Drescher et al., 2016). The area cultivatedwith oil palm grew faster than the area cultivated with rubberplantations between 1990 and 2011 (Clough et al., 2016).

For this study, we used two data sets of different plot sizes.For the first data set, we delineated 28 large plots (rangingfrom 4 to 84 km2) of 7 different land cover types: forest (FO),rubber (RU), acacia plantation forest (PF), young oil palmplantation (YOP), mature oil palm plantation (MOP), urbanarea (UB) and clear-cut areas (CLC) (Fig. 1). The delineationwas based on visual interpretation in combination with fieldobservations, which were conducted between October andDecember 2013. The large size of the plots was necessaryto make a comparison between MODIS and Landsat im-ages (see Sect. 2.3). For the second data set, we selected 49smaller plots within and outside these 28 large plots (between50 m× 50 m and 1000 m× 1000 m) (Fig. 1), which allowedus to increase the number of plots to use when analyzingLandsat images. These small plots were used to extract theLST, NDVI, albedo (α) and ET from a high-resolution Land-sat satellite image (see section satellite data) for the whichdifferent land cover types of interest.

2.2 Meteorological data

Air temperature and relative air humidity were measured atfour reference meteorological stations located in open areaswithin the study area (Drescher et al., 2016), with thermo-hygrometers (type 1.1025.55.000, Thies Clima, Göttingen,Germany) placed at 2 m height. Measurements were recordedevery 15 s and then averaged and stored in a DL16 Prodata logger (Thies Clima, Göttingen, Germany) as 10 minmean, from February 2013 to December 2015. We used theair temperature from the meteorological stations to compareto MODIS air temperatures (MOD07_L2). The relative airhumidity was used as an input parameter for NASA’s on-line atmospheric correction (ATCOR) parameter tool to de-rive parameters to correct Landsat thermal band for atmo-spheric effects (see Sect. 2.3). We also used air temperature

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017

4622 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

Figure 1. Geographic location of the study area. Jambi province on the Sumatran Island of Indonesia (a, b). The background of the map(c) is a digital elevation model, showing that the plots are located in the lowlands of the Jambi province. The large rectangles are the 28different land cover types; the small squares are the locations of the 49 small plots of the 7 different land cover types. The land cover typesare abbreviated as CLC (clear-cut land), UB (urban area), YOP (young oil palm plantation), MOP (mature oil palm plantation), PF (acaciaplantation forest), RU (rubber plantation) and FO (forest).

and relative humidity (RH) from two eddy covariance fluxtowers located in the study area (Meijide et al., 2017): onein a young oil palm plantation (2 years old; 01◦50.127′ S,103◦17.737′ E) and the other one in a mature oil palm plan-tation (12 years old; 01◦41.584′ S, 103◦23.484′ E). At theseflux towers, air temperature and relative humidity were mea-sured above the canopy with the same instruments as in thereference meteorological stations (see Meijide et al., 2017,for a description of the methodology). At the flux tower lo-cated in the mature oil palm plantation, we also measuredthe surface canopy temperature between August 2014 andDecember 2015, which was compared to MODIS LST es-timates from the same period. The canopy temperature wasmeasured with two infrared sensors (IR100) connected to adata logger (CR3000), both from Campbell Scientific Inc.(Logan, USA). For a regional coverage we used ERA-Interimdaily air temperature grids (http://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/; Dee et al., 2011) from2000 to 2015 at 0.125◦ resolution to study the annual air tem-perature trend in this period.

2.3 Satellite data

A Landsat 7 ETM+ VIS/TIR 30 m resolution surface re-flectance image with low cloud cover, acquired at 10:13(local time) on 19 June 2013 covering the lowland areaof the Jambi province (path 125, row 61), was used in

this study. Like all Landsat 7 ETM+ images acquired af-ter 31 May 2003, the image we used was affected by ascan line error causing a data loss of about 22 % (http://landsat.usgs.gov/products_slcoffbackground.php). Most se-lected plots were located in the center of the image and thusnot affected by the data loss; e.g. the forest plots located atthe edges of the scan line error zone faced minimal data lossbecause they were large enough.

We also downloaded the tile h28v09 of the MODIS Terra(MOD) and Aqua (MYD) daily 1 km Land Surface Temper-ature and Emissivity products (MOD11A1 and MYD11A1Collection 5) and MODIS 16-day 500 m vegetation in-dices NDVI/EVI product (MOD13A1 Collection 5) from5 March 2000 to 31 December 2015 for Terra data and from8 July 2002 to 31 December 2015 for Aqua data. We down-loaded other supporting satellite data such as the MODISAtmospheric Profile product (MOD07_L2) and the MODISGeolocation product (MOD03). All MODIS data were re-projected to WGS84, UTM zone 48 south with the MODISReprojection Tool (MRT). The quality of the MODIS datawas examined with the provided quality flags and only pix-els with the highest-quality flag were used in the analysis.

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4623

2.4 Retrieval of biophysical variables from Landsat 7ETM+ VIS/TIR images

NDVI

NDVI was derived from the reflectances corrected for atmo-spheric effects in the red (ρRED, band 3 Landsat 7 ETM+)and near-infrared (ρNIR, band 4 Landsat 7 ETM+) bands,with

NDVI=ρNIR− ρREDρNIR+ ρRED

. (1)

Surface albedo

The surface albedo (α) was computed with the equation ofLiang (2000) for estimating broadband albedo from Landsatsurface reflectance bands, with

α = 0.3141ρ1+ 0.1607ρ3+ 0.369ρ4

+ 0.1160ρ5+ 0.0456ρ7− 0.0057, (2)

where ρ1, ρ3, ρ4, ρ5 and ρ7 are the Landsat 7 ETM+ surfacereflectance bands (corrected for atmospheric effects).

Land surface temperature

LST was derived following the method proposed by Basti-aanssen (2000), Bastiaanssen et al. (1998a), Coll et al. (2010)and Wukelic et al. (1989) for computing the surface tem-perature from the TIR band (band 6) of Landsat (Supple-ment S1). The TIR band was first converted to thermal ra-diance (L6, W m−2 sr−1 µm−1) and then to atmosphericallycorrected thermal radiance (Rc, W m−2 sr−1 µm−1), as de-scribed by Wukelic et al. (1989) and Coll et al. (2010), andwith the atmospheric parameters obtained on NASA’s onlineAtmospheric Correction Calculator (Barsi et al., 2003, 2005)(Supplement S2). The LST (K) was computed with the fol-lowing equation similar to the Planck equation, as in Coll etal. (2010) and Wukelic et al. (1989):

LST=k2

ln(εNB×k1Rc+ 1

) , (3)

where εNB is the emissivity of the surface ob-tained from the NDVI (Supplement Table S1), k1(= 666.09 mW cm−2 sr−1 µm−1) and k2 (= 1282.71 K)are sensor constants for converting the thermal radianceobtained from band 6 of Landsat 7 to surface temperature.

The surface temperature derived from Landsat thermalband was compared with the MODIS LST product that wasacquired on the same day at 10:30 local time. The LandsatLST image was first resampled to MODIS resolution to en-able a pixel-to-pixel comparison, followed by extracting theaverage LST of 7 land cover types with the data set contain-ing the large delineated plots (Fig. 1).

Evapotranspiration

Based on the Surface Energy Balance Algorithm for Land(SEBAL) (Bastiaanssen, 2000; Bastiaanssen et al., 1998a,b) we estimated ET (mm h−1) from latent heat fluxes (LE,W m−2), which were computed as the residual from sensible(W m−2) and ground (W m−2) heat fluxes subtracted fromnet radiation (Rn, W m−2):

LE = Rn−G−H. (4)

We calculated Rn as the sum of incoming shortwave andlongwave radiation minus the reflected shortwave and long-wave radiation and the emitted longwave radiation (Eq. 5).The surface albedo, surface emissivity and surface tempera-ture determine the amounts of incoming and reflected radia-tion:

Rn = (1−α)Sd ↓ +εaσT4a − (1− ε0)εaσT

4a

− ε0σLST4 , (5)

where Sd ↓ is the incoming shortwave solar radia-tion (W m−2) at the surface, α is the surface albedo(Eq. 2); ε0 is the surface emissivity (−), εa is the atmo-spheric emissivity (−), σ is the Stefan–Boltzmann constant(5.67× 10−8 W m−2 K−4), LST is the surface temperature(K, Eq. 3) and Ta is the sky temperature (K). The surfaceemissivity (ε0) is derived from the NDVI and is described inthe Supplement (Table S1). The average atmospheric emis-sivity (εa) is estimated with the model of Idso and Jackson(1969):

εa = 1− 0.26 × exp((−7.77 × 10−4)

× (273.15− Ta)2). (6)

Ground heat fluxes (W m−2) were derived as a fraction ofRn from an empirical relationship between LST, α and NDVI(Bastiaanssen, 2000) as

G= Rn×LST − 273.15

α×

(0.0038α+ 0.0074α2

)× (1− 0.98NDVI4). (7)

In SEBAL sensible heat (W m−2) was calculated as

H= ρCp1T

rah= ρCp

aLST + brah

, (8)

where ρ is the air density (1.16 kg m−3); Cp is the specificheat of air at constant pressure (1004 J kg−1 K−1); rah is theaerodynamic resistance to heat transport (s m−1); and a and bare regression coefficients, which are determined by a hot ex-treme pixel (where LE= 0 and H is maximum) and a cold ex-treme pixel (where H= 0 and LE is maximum). The aerody-namic resistance to heat transport, rah, is calculated throughan iterative process with air temperature measured at 2 m asinput. SEBAL is described in Bastiaanssen (2000) and Bas-tiaanssen et al. (1998a, b). The application of SEBAL in thisresearch is briefly described in the Supplement S3.

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017

4624 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

2.5 Local short-term differences between differentland cover types

From the created LST, NDVI, albedo and ET images we ex-tracted the average values of the different land cover classeswith the data set containing the small 49 delineated plots cov-ering 7 different land cover types (Fig. 1). The average effectof land transformation, i.e. the change from forest to anothernon-forest land cover type, on the surface temperature wasevaluated as (see Li et al., 2015)

1LST = LSTnon−forest−LSTforest. (9)

A negative 1LST indicates a cooling effect and positive1LST indicates a warming effect of the non-forest vegeta-tion compared with forest. The same procedure was appliedin evaluating the effect of land transformation on the NDVI,albedo and ET.

2.6 Effects of land cover change on the provincialsurface temperature in the past decades

To analyze the long-term effects on the provincial scale weused the MODIS daily LST time series (MOD11A1 andMYD11A1) from 2000 to 2015. MOD11A1 provides LSTfor 10:30 and 22:30 and we used the times series between2000 and 2015. MYD11A1 provides LST for 01:30 and13:30 and is available from 8 July 2002; we used completeyears in our analysis and therefore used the MYD11A1 timeseries from 2003 to 2015. We calculated the mean annualLST at four different times of the day (10:30, 13:30, 22:30and 01:30) between 2000 and 2015 for the lowland of Jambifrom the MODIS daily LST time series (MOD11A1 andMYD11A1). First, we calculated for each pixel the averageLST pixel value using only the best-quality pixels for everyyear. Then, from these pixels we made a composite image(n= 16, one for each year) for the province. Finally, fromeach composite image we calculated the mean annual low-land provincial temperature as the average of all the pixelsthat are enclosed by a zone delineating the lowland of theJambi province. We performed the same analysis with theMODIS 16-day NDVI product (2000–2015) and the ERAdaily temperature grid (2000–2015) to compare the annualtrends of LST, NDVI and air temperature of the province.The average provincial LST and NDVI were compared withthe mean LST and NDVI of a selected forest that remainedundisturbed forest during the 2000–2015 period.

2.7 Statistical analysis

For a comparison of the Landsat-derived LST and theMODIS LST we analyzed the statistical relationships withthe coefficient of determination (R2), the root mean squareerror (RMSE), the mean absolute error (MAE) and the bias:

RMSE=

√√√√√ N∑i=1(Ei −Oi)

2

N, (10)

Bias=

N∑i=1(Ei −Oi)

N, (11)

MAE=

N∑i=1|Ei −Oi |

N, (12)

where Oi is MODIS LST, Ei is the Landsat surface temper-ature and N is the number of pixels compared. Model type 2linear regression was applied for fitting the relation betweenMODIS LST and Landsat LST.

We tested the relation between the biophysical variablesLST (or L6 and Rc, both as pre- or intermediate productsbefore obtaining LST), albedo (α), NDVI and ET with a cor-relation analysis and a multiple linear regression was appliedto analyze the effects of the biophysical variables on the LST.We used the model LST (or Rc or L6)∼α+NDVI+ETand used R2 and standardized β coefficients to evaluate thestrength of the biophysical variables in predicting the LST.

3 Results

3.1 Landsat LST compared to MODIS LST

Landsat and MODIS images showed similar spatial LST pat-terns (Fig. 2). In both images the relatively hot areas (red)correspond to the known clear-cut areas, urban areas or othersparsely vegetated areas, the relatively cool areas (blue) cor-respond to vegetated areas such as forest, plantation forestsand mature oil palm plantations. The coarse-resolution scaleof MODIS (1000 m for LST) allows a large regional cover-age of the study area but does not allow to retrieve detailedinformation on small patches (smaller than 1 km2). In con-trast, the Landsat 7 image allows a detailed study of patchesthat are small enough (as small as 30× 30 m2) but it is af-fected by the scan line error causing data loss at the edges ofthe image. In both MODIS and Landsat images clouds andcloud shadows were removed and therefore lead to data gapsin the images.

Landsat-derived LST correlated well with MODIS LST(R2= 0.82; p<0.001; Fig. 3) with a RMSE of 1.8 ◦C. The

seven land cover types had distinctive LSTs and the observeddifferences between these land cover types were consistent inboth images. The non-vegetated surfaces (CLC and UB) hadhigher surface temperatures than the vegetated surface types(FO, YOP, MOP, PF and RU). Clear-cut land had the highestsurface temperature of all compared land cover types, fol-lowed by urban areas, whereas the vegetated land cover typeshad lower surface temperatures: LSTCLC (39.7± 2.0 ◦C)

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4625

Figure 2. MODIS LST image (a) compared with Landsat LST image (b). Cloud cover and cloud shadow cover resulted in data gaps (nodata). The difference in acquisition time between the images is 15 min. The square in the MODIS image is the area that is covered by theLandsat tile (path 125, row 61). Both satellite images were acquired on 19 June 2013.

> LSTUB (35.8± 1.3 ◦C) > LSTYOP (31.0± 0.7 ◦C) > LSTPF(30.3± 0.7 ◦C) > LSTMOP (29.0± 0.8 ◦C) > LSTRU(27.8± 0.9 ◦C) > LSTFO (27.6± 1.4 ◦C) (Landsat LST,Fig. 3). The same trend was derived from the MODISimage but with higher surface temperatures, except forCLC: LSTCLC (37.7± 1.8 ◦C) > LSTUB (36.3± 1.6 ◦C)> LSTYOP (31.7± 0.9 ◦C) > LSTMOP (30.7± 0.9 ◦C)> LSTPF (29.9± 0.9 ◦C) > LSTRU (29.6± 0.4 ◦C) > LSTFO(29.2± 0.4 ◦C) (MODIS LST, Fig. 3).

3.2 Local short-term differences between differentland cover types

The 1LST between RU, MOP, PF, YOP, UB and CLC landcover types and FO were all positive, meaning that the otherland cover types were warmer than forests (Fig. 4a andSupplement S4 and S5). RU and MOP were 0.4± 1.5 and0.8± 1.2 ◦C warmer than forest, respectively. PF and YOP

were much warmer than forests (1LSTPF−FO = 2.3± 1.1;1LSTYOP−FO = 6.0± 1.9 ◦C). The largest 1LSTs were be-tween forest and the non-vegetated land cover types, i.e. UB(1LST= 8.5± 2.1 ◦C) and CLC (1LST= 10.9± 2.6 ◦C).The LST differences were significant (p<0.05, post hocTukey’s HSD test), except between RU and FO (p= 0.78,post hoc Tukey’s HSD test; Supplement S6, Tables S6.1 andS6.2).

Similar differences were found for the 1NDVI be-tween forest and the other land covers (Fig. 4b). Thenegative 1NDVI indicates that the non-forest land covertypes had lower NDVI than the forest. 1NDVI be-tween FO and RU, MOP, PF and YOP were small (be-tween −0.01± 0.02 (1NDVIMOP−FO) and −0.12± 0.06(1NDVIYOP−FO)). The largest 1NDVIs were between for-est and the non-vegetated land cover types, i.e. UB and CLC(1NDVI=−0.42± 0.11 and −0.41± 0.08, respectively).

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017

4626 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

Figure 3. Average surface temperature (LST) and standard devia-tion (SD) of seven land cover types derived from a Landsat ther-mal image compared with the mean and SD of MODIS LST. Thedashed line is the theoretical 1 : 1 line, and the solid lines are thelinear model type 2 regression line (black) and the confidence lim-its of the regression line (red). The Landsat and MODIS imageswere acquired on 19 June 2013, at 10:13 and 10:30 local time, re-spectively. Landsat pixels (30 m) were resampled to MODIS pixelresolution (926 m) to make a pixel-to-pixel comparison between thetwo sources possible.

All 1NDVIs were significant (p<0.05, post hoc Tukey’sHSD test).

The difference in albedo (1albedo) between forest andthe other land covers was very small (Fig. 4c), with1albedo values between−0.03± 0.01 (1albedoPF−FO) and0.03± 0.02 (1albedoYOP−FO). These differences were sig-nificant (p<0.05, post hoc Tukey’s HSD test). PF had alower albedo than forest (1albedoPF−FO =−0.03± 0.01),while the other land cover types had a higher albedo thanforest.

All compared land covers had lower ET than for-est. RU, MOP and PF had slightly lower ET thanFO (1ETRU−FO =−0.03± 0.04, 1ETMOP−FO =

−0.03± 0.03, 1ETPF−FO=−0.04± 0.03 mm h−1)(Fig. 4d). YOP, UB and CLC had much lower ET valuesthan forests: 1ETYOP−FO =−0.18± 0.04, 1ETUB−FO =

−0.23± 0.04, 1ETCLC−FO=−0.26± 0.06 mm h−1). The1ETs were significant (p<0.05, post hoc Tukey’s HSDtest). The SEBAL-based LE estimates were within thevariability range of LE measurements from eddy covariancemeasurements under similar meteorological conditions (seeSupplement S3).

Figure 4. Differences (mean±SD) in surface temperature(1LST), normalized difference vegetation index (1NDVI), albedo(1albedo) and evapotranspiration (1ET) between other land cov-ers (RU, MOP, PF, YOP, UB and CLC) and forest (FO) in theJambi province, derived from a Landsat LST image acquired on19 June 2013 at 10:13 local time.

Albedo had a weaker influence on the LST (ρ = 0.25,p<0.05) (Table 1) than NDVI and ET. As the thermal ra-diance band (L6) and the atmospherically corrected thermalband (Rc) were the basis for the LST calculation, the highcorrelation between L6 and NDVI (ρ =−0.87, p<0.05)and between L6 and ET (ρ =−0.98, p<0.05) resulted ina high correlation between LST and NDVI (ρ =−0.88) andbetween LST and ET (ρ =−0.98). The analysis showed thatalbedo, NDVI and ET were all significant predictors of LST(F(3, 41586) = 1× 106, p<0.05). ET was the strongest pre-dictor of LST (standardized β =−1.11, p<0.05). Albedo(standardized β =−0.19, p<0.05) and NDVI (standardizedβ =−0.19, p<0.05) were weaker predictors of LST.

A separate analysis (Table S6.3, Supplement S6) showedthat ET was a strong predictor of LST for each land covertype in this study and that NDVI and albedo were minor pre-dictors of LST.

3.3 Effects of land use change on the provincial surfacetemperature in the past decades

The average annual LST of Jambi was characterized bya fluctuating but increasing trend during daytime (Fig. 5aand b) between 2000 and 2015. The average morningLST (10:30) increased by 0.07 ◦C per year (R2

= 0.59;p<0.001), while the midday afternoon LST (13:30 local

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4627

Table 1. Statistical analysis between biophysical variables (albedo (α), NDVI and ET) and spectral radiance band (L6), corrected thermalband (Rc) and Landsat land surface temperature (LST).

Model ρ R2 β Stand. Model fit F statisticsβ (R2)

L6 – α+NDVI+ET α 0.26 0.05 −2.94 −0.19 F (3, 41 586)=NDVI −0.87 0.10 0.23 0.11 0.99 1.10× 106∗∗∗

ET −0.98 1.13 −4.00 −1.16

Rc – α+NDVI+ET α 0.25 0.05 −4.88 −0.20 F (3, 41 586)=NDVI −0.88 0.04 0.16 0.05 0.99 1.79× 106∗∗∗

ET −0.98 1.00 −6.21 −1.10

LST – α+NDVI+ET α 0.25 0.05 −34.01 −0.19 F (3, 41 586)=NDVI −0.88 0.05 1.30 0.05 0.99 2.3× 106∗∗∗

ET −0.98 1.00 −43.53 −1.11

∗∗∗ p= 2× 10−16. LM is multiple linear regression analysis between LST (or L6 or Rc) and three biophysical variables: albedo (α),NDVI and ET. ρ is the correlation coefficient. R2 is the R square of the components. β is the regression coefficient of the component.Stand. β is the standardized β. Model fit (R2) is the overall model fit of the multiple linear regression.

time) increased by 0.13 ◦C per year (R2= 0.35; p= 0.02)

between 2003 and 2015. While the daytime LST showed aclear increase, the night and evening LST (22:30 and 01:30,Fig. 5c and d) trends showed a small decrease of −0.02 ◦C(R2= 0.29; p= 0.02) and −0.01 ◦C (R2

= 0.05; p= 0.50)per year, respectively. The observed LST trends resulted in atotal LST increase of 1.05 and 1.56 ◦C in the morning (10:30)and afternoon (13:30), respectively, and a total decrease ofthe LST of 0.3 ◦C (22:30) and 0.12 ◦C (01:30) at night overthe period from 2000 to 2015 in Jambi.

To separate the effect of land use change from global cli-mate warming, we used a site constantly covered by forestover that period (from the forest sites we used in this study)as a reference that was not directly affected by land coverchanges. That site showed small changes in LST than the en-tire province: only the mean morning LST (10:30) had a sig-nificant but small trend with an increase of 0.03 ◦C per year(R2= 0.21, p<0.05) resulting in a total LST increase of

0.45 ◦C between 2000 and 2015 (Fig. 5a). This LST warmingis much smaller than the overall warming at provincial levelof 1.05 ◦C. The LST time series at other times showed nosignificant trends: the mean afternoon LST (13:30) increasedby −0.05 ◦C per year (R2

= 0.01, p= 0.31) (Fig. 5b), andthe night and evening LST by 0.01◦C per year (Fig. 5c andd, p= 0.19 and p= 0.60, respectively).

The mean annual NDVI in Jambi decreased by 0.002 peryear, resulting in a total NDVI decrease of 0.03 (R2

= 0.34;p= 0.01; Fig. 5e). The NDVI of the forest showed a smallbut not significant increase of 0.001 per year (R2

= 0.04,p= 0.23) (Fig. 5e) fluctuating around an NDVI of 0.84.

The mean annual midday air temperature (at 13:00, lo-cal time, Fig. 5f) and the mean annual night air temper-ature (at 01:00, local time) increased every year by 0.05and 0.03 ◦C, respectively, resulting in a total air tempera-

ture increase of 0.75 ◦C (R2= 0.66, p < 0.001) and 0.45 ◦C

(R2= 0.32, p= 0.01) between 2000 and 2015 (Fig. 5f).

4 Discussion

4.1 Landsat LST compared to MODIS LST

In our study we retrieved the surface temperature from aLandsat image and compared this with MODIS LST. Our re-sults showed a good agreement between both LSTs (Fig. 3),which is comparable to other studies and thus gives con-fidence in our analysis. Bindhu et al. (2013) found also aclose relationship between MODIS LST and Landsat LSTby using the same aggregation resampling technique as ourmethod and found a R2 of 0.90, a slope of 0.90 and anintercept of 25.8 ◦C for LST, compared to our R2 of 0.8,slope of 1.35 and intercept of −11.58 ◦C (Fig. 3). Zhang andHe (2013) validated Landsat LST with MODIS LST and alsofound good agreements (RMSD 0.71–1.87 ◦C) between thetwo sensors, whereas we found a RMSE of 1.71 ◦C. Nev-ertheless, there still are differences and slope versatility be-tween the two satellite sources. These differences are typi-cally caused by differences between the MODIS and Landsatsensors in terms of (a) different sensor properties e.g. spatialand radiometric resolution and sensor calibration; (b) geo-referencing and differences in atmospheric corrections (Li etal., 2004); and (c) emissivity corrections; i.e. the use of ap-proximate equations to derive the emissivity from the NDVIfrom Landsat’s red and NIR bands. Li et al. (2004) andVlassova et al. (2014) identified these same factors in theircomparison of ASTER LST with MODIS LST and LandsatLST with MODIS LST, respectively. Vlassova et al. (2014)found good agreements between MODIS and Landsat LSTand obtained higher LSTs with MODIS than with Land-

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017

4628 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

Figure 5. Mean annual LST (a–d), mean annual NDVI (e) and mean annual air temperature trends (f) in the Jambi province between 2000 and2015 derived from MODIS LST (a 10:30, b 13:30, c 22:30 and d 01:30, local time), MODIS NDVI and ERA-Interim daily air temperature(01:00 and 13:00, local time) data sets, respectively. Grey-shaded areas are the confidence intervals of the means; blue-shaded areas are theconfidence intervals of the regression lines. MODIS LST time series for 13:30 and 01:30 were available from the middle of 2002; for thisreason we used the complete years from 2003 to 2015.

sat, which they attributed to the delay of 15 min in acqui-sition time between MODIS and Landsat. MODIS LST ismeasured 15 min later and our results showed that MODISLSTs were indeed higher than Landsat LST. A comparisonof MODIS LST with locally measured canopy surface tem-peratures during the overpass time of MODIS also showedagreement (Supplement S7, Fig. S7.1). The slope was pos-sibly related to differences in instrumentation and emissivitycorrections and to scale issues; this comparison could cor-roborate the quality check of MODIS LST.

As the MODIS LST product is proven to be accuratewithin 1 ◦C (Silvério et al., 2015; Wan et al., 2004) and has

been intensively validated, the use of MODIS LST was aproper way to assess the quality of our Landsat LST.

The errors from the different sources (such as atmo-spheric correction, emissivity correction, resampling Land-sat to MODIS resolution) are difficult to quantify. Whenwe tested the impact of atmospheric correction and emissiv-ity errors on the LST from Landsat retrieval we found that(a) the overall patterns across different land use types didnot change, (b) emissivity was the most important factor, al-though the effects on LST retrieval were small, and (c) er-rors related to atmospheric correction parameters were smallbecause there were minor differences between the default at-

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4629

mospheric correction (ATCOR) parameters and the ATCORparameters derived with actual local conditions (RH, air pres-sure and air temperature). Following the method of Coll etal. (2009) and Jiang et al. (2015) we show that the use ofthe online atmospheric correction parameter calculator is agood option provided that RH, air temperature and air pres-sure measurements are available. We additionally comparedlocally measured air temperatures with MODIS air tempera-ture and found a good agreement (Supplement S8, Fig. S8.1),which served as a verification that we used a correct air tem-perature for the atmospheric correction parameter calculator.

Overall, our comparison of Landsat LST with MODISLST and against ground observations suggests that we areable to retrieve meaningful spatial and temporal patterns ofLST in the Jambi province.

4.2 LST patterns across different land use and landcover (LULC) types

The land cover types in our study covered a range of landsurface types that develop after forest conversion. This isthe first study in this region that includes oil palm and rub-ber as land use types that develop after forest conversion.The coolest temperatures were at the vegetated land covertypes while the warmest surface temperatures were on thenon-vegetated surface types like urban areas and bare land.Interestingly, the oil palm and rubber plantations were onlyslightly warmer than the forests whereas the young oil palmplantations had clearly higher LST than the other vegetatedsurfaces. For other parts of the world, Lim et al. (2005, 2008),Fall et al. (2010) and Weng et al. (2004) also observed coolertemperatures for forests and the highest surface temperaturesfor barren and urban areas.

In Indonesia, land transformation is often not instanta-neous from forest to oil palm or rubber plantation but canbe associated with several years of bare or abandoned landin-between (Sheil et al., 2009). Oil palm plantations typi-cally have a rotation cycle of 25 years, resulting in repeatingpatterns with young plantations (Dislich et al., 2016). Giventhe large LST differences between forests and bare soils oryoung oil palm plantations that we observed, a substantialwarming effect of land transformation at regional scale is ex-pected.

4.3 Drivers of local differences between differentland cover types

All the land cover types (except acacia plantation forests)had a higher albedo than forest, indicating that these landcover types absorbed less incoming solar radiation thanforests. Nevertheless, these land cover types were warmerthan forests, suggesting that the albedo was not the dominantvariable explaining the LST. Indeed, the statistical analysisshowed that ET–LST had a higher correlation than albedo–LST. The 1ETs were significant, highlighting that despite

their higher albedo, all land cover types had higher LSTsthan forests related to lower ET rates than forests. In con-trast, forests that absorb more solar radiation because of thelower albedo have lower LST because of the higher ET theyexhibit, hereby identifying evaporative cooling as the maindeterminant of regulating the surface temperature of all veg-etation cover types (Li et al., 2015).

Both observational and modeling studies carried out inother geographic regions and with other trajectories supportour observations. Observational studies in the Amazonia byLawrence and Vandecar (2015) on the conversion of naturalvegetation to crop or pasture land showed a surface warmingeffect. Salazar et al. (2015) provided additional evidence thatconversion of forest to other types of land use in the Ama-zonia caused significant reductions in precipitation and in-creases in surface temperatures.

Alkama and Cescatti (2016) and earlier studies by Loarieet al. (2011a, b) showed that tropical deforestation may in-crease the LST. Croplands in the Amazonian regions werealso warmer than forests through the reduction of ET (Ban-Weiss et al., 2011; Feddema et al., 2005) and that the climaticresponse strongly depends on changes in energy fluxes ratherthan on albedo changes (Loarie et al., 2011a, b). A study bySilvério et al. (2015) indeed found that tropical deforesta-tion changes the surface energy balance and water cycle andthat the magnitude of the change strongly depends on theland uses that follow deforestation. They found that the LSTwas 6.4 ◦C higher over croplands and 4.3 ◦C higher over pas-ture lands compared to the forest they replaced, as a conse-quence of energy balance shifts. Ban-Weiss et al. (2011) andDavin and de Noblet-Ducoudré (2010) added that in additionto the reduction of ET, the reduction of surface roughnessmost likely enhanced the substantial local warming.

Also for non-Amazonian regions, the replacement offorests by crops caused changes comparable with our obser-vations. In temperate Argentina, Houspanossian et al. (2013)found that the replacement of dry forests by crops resultedin an increase of albedo but still forests exhibited coolercanopies than croplands. The cooler canopies were a resultof a higher aerodynamic conductance that enhanced the ca-pacity of tree canopies to dissipate heat into the atmosphereand to both latent and sensible heat fluxes operating simulta-neously to cool forest canopies.

In a global analysis Li et al. (2015) showed that tropicalforests generally have a low albedo, but still the net energygain caused by solar energy absorption is offset by a greaterlatent heat loss via higher ET and that in the tropical foreststhe high ET cooling completely offsets the albedo warming.For China, this cooling effect was also shown by Peng etal. (2014), who compared LST, albedo and ET of plantationforests, grassland and cropland with forests.

Using NDVI as an indicator of vegetation abundanceWeng et al. (2004) (for the US) and Yue et al. (2007) (forChina) found that areas with a high mean NDVI had a lowerLST than areas with a low mean NDVI, therefore suggesting

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017

4630 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

that vegetation abundance is an important factor in control-ling the LST through higher ET rates. Our result support theirassumptions by showing the high correlation between NDVIand LST and between ET and LST.

Our findings are also supported by modeling studies.Beltrán-Przekurat et al. (2012) found for the southern Ama-zon that conversion of wooded vegetation to soybean planta-tions caused an increase of the LST due to decreased latentheat and increased sensible heat fluxes. Climate models alsoshow the same warming trends and land surface modelingalso projects an increase in surface temperatures followingdeforestation in the Brazilian Cerrado (Beltrán-Przekurat etal., 2012; Loarie et al., 2011b). In a global analysis, Pon-gratz et al. (2006) showed a LST increase of forest to crop-land or pasture transitions, which was driven by a reducedroughness length and an increased aerodynamic resistance,and that the temperature response is intensified in forest toclear/bare land transitions (1.2–1.7 ◦C increase). Similar toobservational studies, the modeling results of Bathiany etal. (2010) show that ET is the main driver of temperaturechanges in tropical land areas.

To understand the effects of deforestation on biophysi-cal variables in Indonesia, our study identifies the followingmechanisms: (a) reduction of ET decreases surface cooling,(b) reduced surface roughness reduces air mixing in the sur-face layer and thus vertical heat fluxes, (c) changes in albedochange the net radiation and (d) changes in energy partition-ing in sensible and latent heat and heat storage. The effect isan increase of the mean temperatures that leads to warmingeffects in all tropical climatic zones (Alkama and Cescatti,2016). We point here that our study included a ground heatflux but did not take into account the storage of heat in thesoil and the release of stored heat out of the soil during thedaily cycle, and the Landsat satellite image was obtained un-der cloud-free conditions with high shortwave radiation inputand low fraction of diffuse radiation. Therefore, the LST re-trieved on cloud-free days might be overestimated comparedto cloudy days, as the differences in LST between land usesare supposed to be lower when diffuse radiation increases.

Our study is the first to include the oil palm and rubber ex-pansion in Indonesia. In Indonesia, smallholders take 40 %of the land under oil palm cultivation for their account (Dis-lich et al., 2016). Because the landscape in Jambi is charac-terized by a small-scale smallholder-dominated mosaic, in-cluding rubber and oil palm monocultures (Clough et al.,2016), studies using medium- to coarse-resolution data arenot able to capture the small-scale changes and processes atthe small-scale level. By using high-resolution Landsat datawe were also able to include the effects of land use changeon biophysical variables and the underlying processes of thesmall-scale holder agriculture.

4.4 Effects of land use change on the provincial surfacetemperature in the past decades

The increases of the mean surface temperature in Jambi werestronger during the morning (10:30) and afternoon (13:30)than during the evening (22:30) and night (01:30). Given thatour results show a decrease of the NDVI in the same period,this suggests that the observed increased trend of the daytime LST can be attributed to the land cover changes that oc-curred. Our assumption that the observed decreasing NDVItrend is caused by land conversions is supported by two dif-ferent studies which reported that in Jambi, between 2000and 2011 (Drescher et al., 2016) and between 2000 and 2013(Clough et al., 2016), the forest area decreased and that thelargest increases were for rubber, oil palm and agriculturaland tree crop areas. The class “other land use types”, whichincludes urban areas, showed a minor increase (around 1 %),suggesting that the decrease in NDVI was most likely causedby forest cover loss and not by urban expansion (Table S9).The same observations on LULC change in Indonesia werealso done by Lee et al. (2011), Margono et al. (2012, 2014)and Luskin et al. (2014). Luskin et al. (2014) showed thatin Jambi, during the period 2000–2010, forests decreased by17 % while oil palm and rubber area increased by 85 and19 %, respectively.

Given these trends in LULC changes, the observed LSTtrends were most likely caused by gradual decrease offorest cover loss at the expense of agriculture and crop-lands. Our assumptions are supported by findings of Sil-vério et al. (2015), Costa et al. (2007), Oliveira et al. (2013),Spracklen et al. (2012) and Salazar et al. (2015) that in-dicate that land use transitions in deforested areas likelyhave a strong influence on regional climate. Alkama andCescatti (2016) show that biophysical effects of forest coverchanges can substantially affect the local climate by alteringthe average temperature, which is consistent with our obser-vations and can be related to the observed land use change inthe Jambi province. As Indonesia has undergone high rates offorest cover loss from 2000 to 2012 (Margono et al., 2014),these findings support our assumptions that the observed LSTincrease in the Jambi province was most likely caused by theobserved land use changes.

To separate the effect of global warming from land-use-change-induced warming, we considered areas with perma-nent and large enough forests as a reference where changesare mainly because of global warming. We find that LST offorests shows either no significant trends (at 01:30, 22:30,01:30) or just a clearly smaller increase of 0.03 ◦C per year at10:30. The difference between the LST trend of the provinceand of the forest at 10:30 was 0.04 ◦C per year, resulting in a1LST of 0.6 ◦C between the province and forest in the period2000 and 2015. We point out that our MODIS analysis has alarger proportion of data from the dry season compared fromthe wet season, as there were more cloud-free conditions dur-ing the dry season. Thus, our reported warming effect reflects

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4631

cloud-free conditions. During cloudy conditions, particularlyin the wet season, the warming effect is expected to be lower.A seasonality analysis showed that the relationships in thedry season are stronger than for the wet season (see Supple-ment S10, Fig. S10.1), which suggests that the warming ismore pronounced during the dry season compared to the wetseason, which is reasonable as we have more incoming radi-ation during the dry season.

With the warming effects we found between forest andother land cover types (1LST, Fig. 4a) and the observedland cover changes by Clough et al. (2016) and Drescher etal. (2016) (Supplement S9, Tables S9.1 and S9.2), we esti-mated the contribution of all land cover types (except for-est) to the 1LST of the province between 2000 and 2015to be 0.51◦C out of the observed 0.6◦C, which also sup-ports our assumption that the LST increase in Jambi was for85 % driven by land cover changes (see Supplement S9, Ta-bles S9.1 and S9.2: Land use change analysis), with clear-cut areas having a large contribution as they have the largestwarming effect.

The observed small but significant increase in LST offorests of 0.03 ◦C per year at 10:30 reflects a LST change in-dependent of land cover changes, as the forest remained un-changed over that time period. A potential driver of that LSTincrease is the general global air temperature trend becauseof changes in radiative forcing or border effects (advectionfrom warmer land uses), which is similar to the 1994–2014time series analysis of Kayet et al. (2016), who showed aLST increase for all land cover types ranging from wastedland, agriculture land, open forest, dense forest, water bodiesand built-up areas.

The observed trends of the provincial air temperature(Fig. 5f) were significant, suggesting that a general warmingdue to global and regional effects contributes to the observedwarming at the provincial level during day and nighttime, butthat it is smaller than the land-cover-change-induced effects(Supplement S9, Tables S9.1 and S9.2) at the provincial level(Fig. 5a and b).

In our long-term analysis on the regional effects of landuse change we observed an increase in the mean LST andmean air temperature in the 2000–2015 period, concurrentwith a decrease of the NDVI. The warming observed fromMODIS LST data and from the air temperature obtainedfrom the independent ERA-Interim reanalysis in the Jambiprovince are most likely caused by the observed decrease ofthe forest area and an increase of oil palm, rubber and othercash crop areas in the same period, with other effects such asradiative forcing changes and additional natural effects play-ing a smaller role. Given the plan of the Indonesian govern-ment to substantially expand oil palm production with a pro-jected additional demand of 1 to 28 Mha in 2020 (Wicke etal., 2011), the strong warming effect we show for Jambi mayserve as an indication of future LST changes for other regionsof Indonesia that will undergo land transformations towardsoil palm plantations.

A recent study by Tölle et al. (2017) showed that forSoutheast Asia, land use change at large scale may not onlyincrease surface temperature but also impact other aspects oflocal and regional weather and climate, including in regionsremote from the original landscape disturbance. Their resultsalso indicate that land clearings can amplify the response toclimatic extreme events such as El Niño–Southern Oscilla-tion. The observed effects of land use change on the biophys-ical variables may have implications for ecosystem servicesin the Jambi province beyond a pure warming effect. Thehigh precipitation in this region in combination with the re-duced vegetation cover of bare land and young oil palm plan-tations impose risks of soil erosion caused by surface run off.Less water infiltration into the soil, thereby decreasing thesoil water storage, may lead to low water availability in thedry season (Dislich et al., 2016; Merten et al., 2016). Highsurface temperatures in combination with low water avail-ability may make the vegetation and the surroundings morevulnerable to fires.

5 Conclusions

In summary, we studied the effects of land use and land coverchanges on the surface biophysical variables in Jambi andexplained the underlying mechanisms of the surface tem-perature regulation. We derived biophysical variables fromsatellite data, analyzed the biophysical impacts of deforesta-tion and on a local scale we found a general warming effectafter forests are transformed to cash or tree croplands (oilpalm, rubber, acacia) in the Jambi province of Sumatra. Thewarming effect after forest conversion results from the re-duced evaporative cooling, which was identified as the maindeterminant of regulating the surface temperature. On a re-gional scale, we saw that the effects of land cover changesare reflected back in changes of the LST, NDVI and air tem-perature in Jambi. The warming effect induced by land coverchange clearly exceeded the global warming effect. Under-standing the effects of land cover change on the biophysicalvariables may support policies regarding conservation of theexisting forests, planning and expansion of the oil palm plan-tations and possible afforestation measures.

Data availability. Data are available upon request from the corre-sponding author. MODIS and Landsat satellite data are distributedby the Land Processes Distributed Active Archive Center (LPDAAC), located at USGS/Earth Resources Observation and Sci-ence USGS/EROS Centre, Sioux Falls, SD (http://lpdaac.usgs.gov),and are freely available and accessible. ERA-Interim data areprovided by European Centre for Medium-Range Weather Fore-casts (ECMWF) and available at http://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/.

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017

4632 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

The Supplement related to this article is available onlineat https://doi.org/10.5194/bg-14-4619-2017-supplement.

Author contributions. CRS conducted the research, fieldwork andanalysis and prepared the manuscript, which was reviewed by GlM,TJ, AM, OR and AK. AM and AK provided the meteorologicaldata.

Competing interests. The authors declare that they have no conflictof interest.

Acknowledgements. This research was funded by the ErasmusMundus Joint Doctorate Programme Forest and Nature for Society(EMJD FONASO) and the German Research Foundation (DFG)through the CRC 990 “EFForTS, Ecological and SocioeconomicFunctions of Tropical Lowland Rainforest Transformation Systems(Sumatra, Indonesia)” (subproject A03). A special thanks toHuta Julu Bagus Putra, a.k.a. Monang, for his assistance andtranslation during the field work in Indonesia.

This open-access publication was fundedby the University of Göttingen.

Edited by: Paul StoyReviewed by: two anonymous referees

References

Alkama, R. and Cescatti, A.: Biophysical climate impacts of re-cent changes in global forest cover, Science, 351, 600–604,https://doi.org/10.1126/science.aac8083, 2016.

Ban-Weiss, G. A., Bala, G., Cao, L., Pongratz, J., and Caldeira,K.: Climate forcing and response to idealized changes in sur-face latent and sensible heat, Environ. Res. Lett., 6, 34032,https://doi.org/10.1088/1748-9326/6/3/034032, 2011.

Barsi, J. A., Barker, J. L., and Schott, J. R.: An Atmo-spheric Correction Parameter Calculator for a Single Ther-mal Band Earth-Sensing Instrument, Geosci. Remote Sens.Symp. 2003 IGARSS 03 Proc. 2003 Institute of Electri-cal and Electronics Engineers International, 5, 3014–3016,https://doi.org/10.1109/IGARSS.2003.1294665, 2003.

Barsi, J. A., Schott, J. R., Palluconi, F. D., and Hook, S. J.:Validation of a web-based atmospheric correction tool forsingle thermal band instruments, in: Proc. SPIE, Earth Ob-serving Systems X, San Diego, California, USA, Vol. 5882,https://doi.org/10.1117/12.619990, 2005.

Bastiaanssen, W. G. M., Menenti, M., Feddes, R. A., and Holtslag,A. A. M.: A remote sensing surface energy balance algorithmfor land (SEBAL) – 1. Formulation, J. Hydrol., 212, 198–212,https://doi.org/10.1016/s0022-1694(98)00253-4, 1998a.

Bastiaanssen, W. G. M., Pelgrum, H., Wang, J., Ma, Y., Moreno, J.F., Roerink, G. J., and van der Wal, T.: A remote sensing surfaceenergy balance algorithm for land (SEBAL) – Part 2: Validation,

J. Hydrol., 212–213, 213–229, https://doi.org/10.1016/S0022-1694(98)00254-6, 1998b.

Bastiaanssen, W. G.: SEBAL-based sensible and latent heat fluxesin the irrigated Gediz Basin, Turkey, J. Hydrol., 229, 87–100,https://doi.org/10.1016/S0022-1694(99)00202-4, 2000.

Bathiany, S., Claussen, M., Brovkin, V., Raddatz, T., and Gayler, V.:Combined biogeophysical and biogeochemical effects of large-scale forest cover changes in the MPI earth system model, Bio-geosciences, 7, 1383–1399, https://doi.org/10.5194/bg-7-1383-2010, 2010.

Beltrán-Przekurat, A., Pielke Sr, R. A., Eastman, J. L.,and Coughenour, M. B.: Modelling the effects of land-use/land-cover changes on the near-surface atmosphere insouthern South America, Int. J. Climatol., 32, 1206–1225,https://doi.org/10.1002/joc.2346, 2012.

Bindhu, V. M., Narasimhan, B., and Sudheer, K. P.: Develop-ment and verification of a non-linear disaggregation method(NL-DisTrad) to downscale MODIS land surface tempera-ture to the spatial scale of Landsat thermal data to esti-mate evapotranspiration, Remote Sens. Environ., 135, 118–129,https://doi.org/10.1016/j.rse.2013.03.023, 2013.

Boisier, J. P., de Noblet-Ducoudré, N., and Ciais, P.: Historical land-use-induced evapotranspiration changes estimated from present-day observations and reconstructed land-cover maps, Hydrol.Earth Syst. Sci., 18, 3571–3590, https://doi.org/10.5194/hess-18-3571-2014, 2014.

Bridhikitti, A. and Overcamp, T. J.: Estimation of Southeast Asianrice paddy areas with different ecosystems from moderate-resolution satellite imagery, Agric. Ecosyst. Environ., 146, 113–120, https://doi.org/10.1016/j.agee.2011.10.016, 2012.

Bright, R. M., Zhao, K., Jackson, R. B., and Cherubini, F.: Quantify-ing surface albedo and other direct biogeophysical climate forc-ings of forestry activities, Glob. Change Biol., 21, 3246–3266,https://doi.org/10.1111/gcb.12951, 2015.

Clough, Y., Krishna, V. V., Corre, M. D., Darras, K., Denmead, L.H., Meijide, A., Moser, S., Musshoff, O., Steinebach, S., Veld-kamp, E., Allen, K., Barnes, A. D., Breidenbach, N., Brose, U.,Buchori, D., Daniel, R., Finkeldey, R., Harahap, I., Hertel, D.,Holtkamp, A. M., Hörandl, E., Irawan, B., Jaya, I. N. S., Jochum,M., Klarner, B., Knohl, A., Kotowska, M. M., Krashevska, V.,Kreft, H., Kurniawan, S., Leuschner, C., Maraun, M., Melati, D.N., Opfermann, N., Pérez-Cruzado, C., Prabowo, W. E., Rem-bold, K., Rizali, A., Rubiana, R., Schneider, D., Tjitrosoedirdjo,S. S., Tjoa, A., Tscharntke, T., and Scheu, S.: Land-use choicesfollow profitability at the expense of ecological functions inIndonesian smallholder landscapes, Nat. Commun., 7, 13137,https://doi.org/10.1038/ncomms13137, 2016.

Coll, C., Wan, Z., and Galve, J. M.: Temperature-based andradiance-based validations of the V5 MODIS land sur-face temperature product, J. Geophys. Res., 114, D20102,https://doi.org/10.1029/2009JD012038, 2009.

Coll, C., Galve, J. M., Sanchez, J. M., and Caselles, V.:Validation of Landsat-7/ETM+ Thermal-Band Calibrationand Atmospheric Correction With Ground-Based Measure-ments, IEEE Trans. Geosci. Remote Sens., 48, 547–555,https://doi.org/10.1109/TGRS.2009.2024934, 2010.

Costa, M. H., Yanagi, S. N. M., Souza, P. J. O. P., Ribeiro,A., and Rocha, E. J. P.: Climate change in Amazonia causedby soybean cropland expansion, as compared to caused by

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4633

pastureland expansion, Geophys. Res. Lett., 34, L07706,https://doi.org/10.1029/2007GL029271, 2007.

Davin, E. L. and de Noblet-Ducoudré, N.: Climatic Im-pact of Global-Scale Deforestation: Radiative ver-sus Nonradiative Processes, J. Clim., 23, 97–112,https://doi.org/10.1175/2009JCLI3102.1, 2010.

Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli,P., Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G.,Bauer, P., Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bid-lot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer,A. J., Haimberger, L., Healy, S. B., Hersbach, H., Hólm, E. V.,Isaksen, L., Kållberg, P., Köhler, M., Matricardi, M., McNally,A. P., Monge-Sanz, B. M., Morcrette, J.-J., Park, B.-K., Peubey,C., de Rosnay, P., Tavolato, C., Thépaut, J.-N., and Vitart, F.: TheERA-Interim reanalysis: configuration and performance of thedata assimilation system, Q. J. R. Meteorol. Soc., 137, 553–597,https://doi.org/10.1002/qj.828, 2011.

Dislich, C., Keyel, A. C., Salecker, J., Kisel, Y., Meyer, K. M.,Auliya, M., Barnes, A. D., Corre, M. D., Darras, K., Faust,H., Hess, B., Klasen, S., Knohl, A., Kreft, H., Meijide, A.,Nurdiansyah, F., Otten, F., Pe’er, G., Steinebach, S., Tarigan,S., Tölle, M. H., Tscharntke, T., and Wiegand, K.: A re-view of the ecosystem functions in oil palm plantations, us-ing forests as a reference system, Biol. Rev., 92, 1539–1569,https://doi.org/10.1111/brv.12295, 2016.

Drescher, J., Rembold, K., Allen, K., Beckschäfer, P., Buchori, D.,Clough, Y., Faust, H., Fauzi, A. M., Gunawan, D., Hertel, D.,Irawan, B., Jaya, I. N. S., Klarner, B., Kleinn, C., Knohl, A.,Kotowska, M. M., Krashevska, V., Krishna, V., Leuschner, C.,Lorenz, W., Meijide, A., Melati, D., Nomura, M., Pérez-Cruzado,C., Qaim, M., Siregar, I. Z., Steinebach, S., Tjoa, A., Tscharntke,T., Wick, B., Wiegand, K., Kreft, H., and Scheu, S.: Ecologicaland socio-economic functions across tropical land use systemsafter rainforest conversion, Philos. Trans. R. Soc. Lond. B, 371,20150275, https://doi.org/10.1098/rstb.2015.0275, 2016.

Fall, S., Niyogi, D., Gluhovsky, A., Pielke, R. A., Kalnay, E., andRochon, G.: Impacts of land use land cover on temperature trendsover the continental United States: assessment using the NorthAmerican Regional Reanalysis, Int. J. Climatol., 30, 1980–1993,https://doi.org/10.1002/joc.1996, 2010.

Feddema, J. J., Oleson, K. W., Bonan, G. B., Mearns, L. O., Buja,L. E., Meehl, G. A., and Washington, W. M.: The Importanceof Land-Cover Change in Simulating Future Climates, Science,310, 1674–1678, https://doi.org/10.1126/science.1118160, 2005.

Hoffmann, W. A. and Jackson, R. B.: Vegetation–Climate Feed-backs in the Conversion of Tropical Savanna to Grass-land, J. Clim., 13, 1593–1602, https://doi.org/10.1175/1520-0442(2000)013<1593:VCFITC>2.0.CO;2, 2000.

Houspanossian, J., Nosetto, M., and Jobbágy, E. G.: Radi-ation budget changes with dry forest clearing in tem-perate Argentina, Glob. Change Biol., 19, 1211–1222,https://doi.org/10.1111/gcb.12121, 2013.

Idso, S. B. and Jackson, R. D.: Thermal radiation fromthe atmosphere, J. Geophys. Res., 74, 5397–5403,https://doi.org/10.1029/JC074i023p05397, 1969.

Jiang, Y., Fu, P. and Weng, Q.: Assessing the Impacts ofUrbanization-Associated Land Use/Cover Change on Land Sur-face Temperature and Surface Moisture: A Case Study in

the Midwestern United States, Remote Sens., 7, 4880–4898,https://doi.org/10.3390/rs70404880, 2015.

Kayet, N., Pathak, K., Chakrabarty, A., and Sahoo, S.: Spatial im-pact of land use/land cover change on surface temperature dis-tribution in Saranda Forest, Jharkhand, Model. Earth Syst. Envi-ron., 2, 1–10, https://doi.org/10.1007/s40808-016-0159-x, 2016.

Lawrence, D. and Vandecar, K.: Effects of tropical deforestationon climate and agriculture, Nature Climate Change, 5, 27–36,https://doi.org/10.1038/nclimate2430, 2015.

Lee, X., Goulden, M. L., Hollinger, D. Y., Barr, A., Black, T. A.,Bohrer, G., Bracho, R., Drake, B., Goldstein, A., Gu, L., Katul,G., Kolb, T., Law, B. E., Margolis, H., Meyers, T., Monson, R.,Munger, W., Oren, R., Paw U, K. T., Richardson, A. D., Schmid,H. P., Staebler, R., Wofsy, S., and Zhao, L.: Observed increase inlocal cooling effect of deforestation at higher latitudes, Nature,479, 384–387, https://doi.org/10.1038/nature10588, 2011.

Li, F., Jackson, T. J., Kustas, W. P., Schmugge, T. J., French, A. N.,Cosh, M. H., and Bindlish, R.: Deriving land surface tempera-ture from Landsat 5 and 7 during SMEX02/SMACEX, 2002 SoilMoisture Exp. SMEX02, Remote Sens. Environ., 92, 521–534,https://doi.org/10.1016/j.rse.2004.02.018, 2004.

Li, Y., Zhao, M., Motesharrei, S., Mu, Q., Kalnay, E.,and Li, S.: Local cooling and warming effects of forestsbased on satellite observations, Nat. Commun., 6, 6603,https://doi.org/10.1038/ncomms7603, 2015.

Liang, S.: Narrowband to broadband conversions of land surfacealbedo I: Algorithms, Remote Sens. Environ., 76, 213–238,https://doi.org/10.1016/S0034-4257(00)00205-4, 2000.

Lim, Y.-K., Cai, M., Kalnay, E., and Zhou, L.: Observationalevidence of sensitivity of surface climate changes to landtypes and urbanization, Geophys. Res. Lett., 32, L22712,https://doi.org/10.1029/2005GL024267, 2005.

Lim, Y.-K., Cai, M., Kalnay, E., and Zhou, L.: Impact of VegetationTypes on Surface Temperature Change, J. Appl. Meteorol. Cli-matol., 47, 411–424, https://doi.org/10.1175/2007JAMC1494.1,2008.

Loarie, S. R., Lobell, D. B., Asner, G. P., Mu, Q., andField, C. B.: Direct impacts on local climate of sugar-caneexpansion in Brazil, Nature Climate Change, 1, 105–109,https://doi.org/10.1038/nclimate1067, 2011a.

Loarie, S. R., Lobell, D. B., Asner, G. P., and Field, C.B.: Land-Cover and Surface Water Change Drive LargeAlbedo Increases in South America, Earth Interact., 15, 1–16,https://doi.org/10.1175/2010EI342.1, 2011b.

Longobardi, P., Montenegro, A., Beltrami, H., andEby, M.: Deforestation Induced Climate Change: Ef-fects of Spatial Scale, PLoS ONE, 11, e0153357,https://doi.org/10.1371/journal.pone.0153357, 2016.

Luskin, M. S., Christina, E. D., Kelley, L. C., and Potts, M. D.:Modern Hunting Practices and Wild Meat Trade in the OilPalm Plantation-Dominated Landscapes of Sumatra, Indone-sia, Hum. Ecol., 42, 35–45, https://doi.org/10.1007/s10745-013-9606-8, 2014.

Mahmood, R., Pielke, R. A., Hubbard, K. G., Niyogi, D., Dirmeyer,P. A., McAlpine, C., Carleton, A. M., Hale, R., Gameda, S.,Beltrán-Przekurat, A., Baker, B., McNider, R., Legates, D.R., Shepherd, M., Du, J., Blanken, P. D., Frauenfeld, O. W.,Nair, U. S., and Fall, S.: Land cover changes and their bio-

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017

4634 C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia

geophysical effects on climate, Int. J. Climatol., 34, 929–953,https://doi.org/10.1002/joc.3736, 2014.

Margono, B. A., Turubanova, S., Zhuravleva, I., Potapov, P.,Tyukavina, A., Baccini, A., Goetz, S., and Hansen, M. C.:Mapping and monitoring deforestation and forest degrada-tion in Sumatra (Indonesia) using Landsat time series datasets from 1990 to 2010, Environ. Res. Lett., 7, 34010,https://doi.org/10.1088/1748-9326/7/3/034010, 2012.

Margono, B. A., Potapov, P. V., Turubanova, S., Stolle, F.,and Hansen, M. C.: Primary forest cover loss in Indone-sia over 2000–2012, Nature Climate Change, 4, 730–735,https://doi.org/10.1038/nclimate2277, 2014.

Marlier, M. E., DeFries, R., Pennington, D., Nelson, E., Or-dway, E. M., Lewis, J., Koplitz, S. N., and Mickley, L.J.: Future fire emissions associated with projected landuse change in Sumatra, Glob. Change Biol., 21, 345–362,https://doi.org/10.1111/gcb.12691, 2015.

Meijide, A., Röll, A., Fan, Y., Herbst, M., Niu, F., Tiedemann, F.,June, T., Rauf, A., Hölscher, D., and Knohl, A.: Controls ofwater and energy fluxes in oil palm plantations: Environmentalvariables and oil palm age, Agr. Forest Meteorol., 239, 71–85,https://doi.org/10.1016/j.agrformet.2017.02.034, 2017.

Merten, J., Röll, A., Guillaume, T., Meijide, A., Tarigan, S.,Agusta, H., Dislich, C., Dittrich, C., Faust, H., Gunawan, D.,Hein, J., Hendrayanto, Knohl, A., Kuzyakov, Y., Wiegand, K.,and Hölscher, D.: Water scarcity and oil palm expansion: so-cial views and environmental processes, Ecol. Soc., 21, 5,https://doi.org/10.5751/ES-08214-210205, 2016.

Miettinen, J., Shi, C., and Liew, S. C.: Deforestation rates in insularSoutheast Asia between 2000 and 2010, Glob. Change Biol., 17,2261–2270, https://doi.org/10.1111/j.1365-2486.2011.02398.x,2011.

Miettinen, J., Hooijer, A., Wang, J., Shi, C., and Liew, S.C.: Peatland degradation and conversion sequences and in-terrelations in Sumatra, Reg. Environ. Change, 12, 729–737,https://doi.org/10.1007/s10113-012-0290-9, 2012.

Mildrexler, D. J., Zhao, M., and Running, S. W.: A global compar-ison between station air temperatures and MODIS land surfacetemperatures reveals the cooling role of forests, J. Geophys. Res.-Biogeo., 116, G03025, https://doi.org/10.1029/2010JG001486,2011.

Nosetto, M. D., Jobbágy, E. G., and Paruelo, J. M.: Land-use change and water losses: the case of grassland afforesta-tion across a soil textural gradient in central Argentina, Glob.Change Biol., 11, 1101–1117, https://doi.org/10.1111/j.1365-2486.2005.00975.x, 2005.

Oliveira, L. J. C., Costa, M. H., Soares-Filho, B. S., andCoe, M. T.: Large-scale expansion of agriculture in Amazo-nia may be a no-win scenario, Environ. Res. Lett., 8, 24021,https://doi.org/10.1088/1748-9326/8/2/024021, 2013.

Peng, S.-S., Piao, S., Zeng, Z., Ciais, P., Zhou, L., Li, L. Z. X., My-neni, R. B., Yin, Y., and Zeng, H.: Afforestation in China coolslocal land surface temperature, Proc. Natl. Acad. Sci. USA, 111,2915–2919, https://doi.org/10.1073/pnas.1315126111, 2014.

Pongratz, J., Bounoua, L., DeFries, R. S., Morton, D. C., Anderson,L. O., Mauser, W., and Klink, C. A.: The Impact of Land CoverChange on Surface Energy and Water Balance in Mato Grosso,Brazil, Earth Interact., 10, 1–17, https://doi.org/10.1175/EI176.1,2006.

Salazar, A., Baldi, G., Hirota, M., Syktus, J., and McAlpine,C.: Land use and land cover change impacts onthe regional climate of non-Amazonian South Amer-ica: A review, Glob. Planet. Change, 128, 103–119,https://doi.org/10.1016/j.gloplacha.2015.02.009, 2015.

Salazar, A., Katzfey, J., Thatcher, M., Syktus, J., Wong, K., andMcAlpine, C.: Deforestation changes land–atmosphere interac-tions across South American biomes, Glob. Planet. Change, 139,97–108, https://doi.org/10.1016/j.gloplacha.2016.01.004, 2016.

Sheil, D., Casson, A., Meijaard, E., Van Noordwjik, M., Gaskell,J., Sunderland-Groves, J., Wertz, K., and Kanninen, M.: Theimpacts and opportunities of oil palm in Southeast Asia: Whatdo we know and what do we need to know?, Center forInternational Forestry Research (CIFOR), Bogor, Indonesia,https://doi.org/10.17528/cifor/002792, 2009.

Silvério, D. V., Brando, P. M., Macedo, M. N., Beck, P. S. A.,Bustamante, M., and Coe, M. T.: Agricultural expansion dom-inates climate changes in southeastern Amazonia: the over-looked non-GHG forcing, Environ. Res. Lett., 10, 104015,https://doi.org/10.1088/1748-9326/10/10/104015, 2015.

Snyder, W. C., Wan, Z., Zhang, Y., and Feng, Y.-Z.: Classification-based emissivity for land surface temperature measure-ment from space, Int. J. Remote Sens., 19, 2753–2774,https://doi.org/10.1080/014311698214497, 1998.

Sobrino, J. A., Jiménez-Muñoz, J. C., and Paolini, L.: Land surfacetemperature retrieval from LANDSAT TM 5, Remote Sens. En-viron., 90, 434–440, https://doi.org/10.1016/j.rse.2004.02.003,2004.

Sobrino, J. A., Jiménez-Muñoz, J. C., Zarco-Tejada, P. J.,Sepulcre-Cantó, G., and de Miguel, E.: Land surface tem-perature derived from airborne hyperspectral scanner ther-mal infrared data, Remote Sens. Environ., 102, 99–115,https://doi.org/10.1016/j.rse.2006.02.001, 2006.

Sobrino, J. A., Jimenez-Muoz, J. C., Soria, G., Romaguera, M.,Guanter, L., Moreno, J., Plaza, A., and Martinez, P.: LandSurface Emissivity Retrieval From Different VNIR and TIRSensors, IEEE Trans. Geosci. Remote Sens., 46, 316–327,https://doi.org/10.1109/TGRS.2007.904834, 2008.

Spracklen, D. V., Arnold, S. R., and Taylor, C. M.: Observations ofincreased tropical rainfall preceded by air passage over forests,Nature, 489, 282–285, https://doi.org/10.1038/nature11390,2012.

Tölle, M. H., Engler, S., and Panitz, H.-J.: Impact of AbruptLand Cover Changes by Tropical Deforestation on South-east Asian Climate and Agriculture, J. Clim., 30, 2587–2600,https://doi.org/10.1175/JCLI-D-16-0131.1, 2017.

van Leeuwen, T. T., Frank, A. J., Jin, Y., Smyth, P., Goulden,M. L., van der Werf, G. R., and Randerson, J. T.: Opti-mal use of land surface temperature data to detect changesin tropical forest cover, J. Geophys. Res., 116, G02002,https://doi.org/10.1029/2010JG001488, 2011.

Verstraeten, W. W., Veroustraete, F., and Feyen, J.: Estimat-ing evapotranspiration of European forests from NOAA-imagery at satellite overpass time: Towards an operationalprocessing chain for integrated optical and thermal sen-sor data products, Remote Sens. Environ., 96, 256–276,https://doi.org/10.1016/j.rse.2005.03.004, 2005.

Vlassova, L., Perez-Cabello, F., Nieto, H., Martín, P., Riaño, D., andde la Riva, J.: Assessment of Methods for Land Surface Temper-

Biogeosciences, 14, 4619–4635, 2017 www.biogeosciences.net/14/4619/2017/

C. R. Sabajo et al.: Surface temperature in the Jambi province in Indonesia 4635

ature Retrieval from Landsat-5 TM Images Applicable to Multi-scale Tree-Grass Ecosystem Modeling, Remote Sens., 6, 4345–4368, https://doi.org/10.3390/rs6054345, 2014.

Voogt, J. A. and Oke, T. R.: Effects of urban surface geometry onremotely-sensed surface temperature, Int. J. Remote Sens., 19,895–920, https://doi.org/10.1080/014311698215784, 1998.

Wan, Z., Zhang, Y., Zhang, Q., and Li, Z.-L.: Quality as-sessment and validation of the MODIS global land sur-face temperature, Int. J. Remote Sens., 25, 261–274,https://doi.org/10.1080/0143116031000116417, 2004.

Wang, J., Pan, F., Soininen, J., Heino, J., and Shen, J.: Nutri-ent enrichment modifies temperature-biodiversity relationshipsin large-scale field experiments, Nat. Commun., 7, 13960,https://doi.org/10.1038/ncomms13960, 2016.

Weng, Q.: Thermal infrared remote sensing for urban cli-mate and environmental studies: Methods, applications, andtrends, ISPRS J. Photogramm. Remote Sens., 64, 335–344,https://doi.org/10.1016/j.isprsjprs.2009.03.007, 2009.

Weng, Q., Lu, D., and Schubring, J.: Estimation of land sur-face temperature–vegetation abundance relationship for urbanheat island studies, Remote Sens. Environ., 89, 467–483,https://doi.org/10.1016/j.rse.2003.11.005, 2004.

Wicke, B., Sikkema, R., Dornburg, V., and Faaij, A.: Explor-ing land use changes and the role of palm oil productionin Indonesia and Malaysia, Land Use Policy, 28, 193–206,https://doi.org/10.1016/j.landusepol.2010.06.001, 2011.

Wukelic, G. E., Gibbons, D. E., Martucci, L. M., and Foote,H. P.: Radiometric calibration of Landsat Thematic Map-per thermal band, Remote Sens. Environ., 28, 339–347,https://doi.org/10.1016/0034-4257(89)90125-9, 1989.

Yue, W., Xu, J., Tan, W., and Xu, L.: The relationship between landsurface temperature and NDVI with remote sensing: applicationto Shanghai Landsat 7 ETM+ data, Int. J. Remote Sens., 28,3205–3226, https://doi.org/10.1080/01431160500306906, 2007.

Zhang, Z. and He, G.: Generation of Landsat surface temperatureproduct for China, 2000–2010, Int. J. Remote Sens., 34, 7369–7375, https://doi.org/10.1080/01431161.2013.820368, 2013.

Zhou, X. and Wang, Y.-C.: Dynamics of Land Surface Temperaturein Response to Land-Use/Cover Change, Geogr. Res., 49, 23–36,https://doi.org/10.1111/j.1745-5871.2010.00686.x, 2011.

www.biogeosciences.net/14/4619/2017/ Biogeosciences, 14, 4619–4635, 2017